§ ÎÐjyãó$—ddlmZmZmZmZd„ZdS)é)Ú IMAGE_PROMPTÚIMAGE_PROMPT_SHORTÚ help_textÚ specialitiescó¸—d}d}g}d|vr|dz }|dz }|dgz }|dvr|dz }|d z }|d gz }td |t|td g¦«}t|Ž|d <|S)Nz’Generate an image from English or Chinese text. The Base and Turbo entries use different sampling defaults; keep the selected checkpoint's preset.z[English/Chinese text-to-image; keep Base/Turbo sampling defaults matched to the checkpoint.ÚcontrolaE ### Structure Guidance Supply a **Control Image** and choose pose, depth, edges or the model's raw control format. The image supplies structure; the text describes the desired appearance. **Control Strength** adjusts adherence. Do not select raw format for an ordinary photo when you intend to extract a pose or depth map.zt Control image + pose/depth/edge/raw process. Prompt describes finished appearance; Control Strength sets adherence.z control image)Úz_image_control2Úz_image_control2_1z¼ **Image Inpainting** uses a source, mask and replacement description for local edits. These Control v2 selections support masks; the original Control entry does not expose this workflow.z* Control v2: source + mask for inpainting.zimage inpaintingzZ-Imagez%https://github.com/Tongyi-MAI/Z-Imager)rrrr)Ú architectureÚinfoÚshortÚ strengthsÚresults úÑØ €MóN)Úmodels.help_sharedrrrrr©rrúrsBðàXÐXÐXÐXÐXÐXÐXÐXÐXÐXÐXÐXðððððr